Sistema neural artificial para identificação de perdas não técnica em consumidores rurais

Abstract

Non-technical losses have a significant impact on power distribution networks, and they are among the major concerns of the agents involved in power systems. Particularly in rural distribution networks, consumers with crop irrigation systems characterize situations of difficult detection of non-technical losses for power utilities, considering different existing irrigation processes, climatic characteristics and difficulties of local inspection. This work presents a proposal for the identification of non-technical energy losses in rural feeders containing pumping systems for irrigation of rice crops. The proposed methodology is based on the correlation of the electric energy consumption patterns, the characteristics of the irrigated area and the climatic conditions of the irrigation period. The methodology uses an Artificial Neural System composed by Artificial Neural Networks, and it uses as inputs: rainfall, temperature, solar incidence, air humidity, installed power load, irrigated area, soil type, soil elevation height, level off automation and irrigation methodology of the rice cultivation. The final results indicate, for each analyzed consumer, the percentage risk of non-technical losses. The Artificial Neural System allows the analysis of irrigated rice crops in any region of Brazil, independently of the characteristics of the crop, soil and environment. The results of the work were obtained and validated from a real data base of harvests from the period between 2009 and 2014, for crops of the State of Rio Grande do Sul.